{"ID":5937783,"CreatedAt":"2026-07-07T03:14:33.014478982Z","UpdatedAt":"2026-07-08T19:22:52.279459246Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.04469","arxiv_id":"2607.04469","title":"Don't Commit Alone: Joint Token Commitment in Diffusion Large Language Models","abstract":"Diffusion large language models (dLLMs) commit multiple tokens per denoising step by decoding each selected position independently from the shared context; when those positions are dependent, the resulting factorization error is captured by conditional total correlation, which confidence-based selection cannot observe from marginals alone. We propose CoCommit, a marker-gated coordination pass that briefly defers commitment: after the usual bundle selection, a learned marker announces the commit set and the backbone's last-$n$ layers are re-applied so marked positions coordinate -- approximating joint-mode decoding -- before greedy argmax writes tokens. The method reuses existing weights with one extra partial forward pass and no auxiliary model. On LLaDA2.1-mini with LoRA adapters and matched greedy inference, joint commitment improves accuracy on all six benchmarks we evaluate, with the largest gains on reasoning and exact-answer tasks.","short_abstract":"Diffusion large language models (dLLMs) commit multiple tokens per denoising step by decoding each selected position independently from the shared context; when those positions are dependent, the resulting factorization error is captured by conditional total correlation, which confidence-based selection cannot observe...","url_abs":"https://arxiv.org/abs/2607.04469","url_pdf":"https://arxiv.org/pdf/2607.04469v1","authors":"[\"Lin Yao\"]","published":"2026-07-05T19:29:13Z","proceeding":"cs.CL","tasks":"[\"cs.CL\"]","methods":"[\"Diffusion Model\",\"Large Language Model\",\"Language Model\",\"LoRA\"]","has_code":false}